Abstract

Geographic Information Systems (GIS) software is used to analyze rainwater harvesting potential in Escambia County, Florida, USA. The approach presented can be replicated using LiDAR data, and the infrared spectrum of National Agriculture Imagery Program (NAIP) imagery. GIS surface maps are analyzed in combination with local utility consumption data to determine potential reductions in potable water consumption for households. The results indicate an extensive urban catchment of rooftop surfaces, and commensurate potential for rainwater harvesting and stormwater attenuation. Sixty two percent of the households analyzed consumed less water than could be potentially harvested. The remaining 38% consumed more water than could be potentially harvested. There are noted and significant differences between the two sample populations, including differences in water consumed and roof size. A comparison of lot size between the two sample populations did not yield any significant difference. The conclusions indicate that the widespread implementation of rainwater harvesting could substantially reduce potable water use in urban areas, and are of use to policy makers, planners, engineers and property owners everywhere.

INTRODUCTION

It is widely acknowledged that the effects of anthropogenic climate change are altering global weather patterns (IPCC 2014). For example, projected increases in tropospheric water vapor are expected to intensify extreme weather events and result in a ‘latitudinal intensification’ and ‘redistribution of global precipitation’. In other words, it is believed that in the future, the wet regions will become wetter, and dry regions will become drier (Marvel & Bonfils 2013). This conclusion is supported by the United States Environmental Protection Agency's Rate of Precipitation Change Indicator, which has accounted for historical changes in regional precipitation between 1901 and 2012 (US EPA 2015). Similarly, the Australian Government's Bureau of Meteorology has produced a series of maps, detailing trends in climate, including rainfall, between 1970 and 2014 (Bureau of Meteorology 2015). Current weather observations also seem to fit this pattern, as ‘exceptional’ drought conditions covered much of California during 2015 (US Drought Monitor 2015), with State snowpack levels at record lows (USDA 2015), and similarly dry conditions were reported in Australia between 2003 and 2012 (CSIRO 2015). The prospect of less predictable weather patterns has prompted some to rethink the provision of urban water services (Grant et al. 2013).

The impacts of urban development on natural hydrological systems are profound and comprehensive (Paul & Meyer 2001). The impacts of the anthropogenic sealing of soil for example, include more reflective surfaces, increased heat island effect, decreased filtration, increased runoff, erosion of adjacent soils and streams, and increased airborne particulate matter (Scalenghe & Marsan 2009). Jennings and Jarnagin have made historical observations of changes in Total Impervious Area (TIA) and the associated impacts on surrounding water bodies. The authors conclude that increased TIA accelerates normal hydrological processes, and leads to statistically significant increases in streamflow discharge (Jennings & Jarnagin 2002). A similar conclusion was reached by Mejia and Moglen in an analysis of the spatial distribution of impervious surfaces (Mejia & Moglen 2010), which is further supported by reductions in interception, infiltration and evapotranspiration as identified by Dougherty et al. (2007). It is widely believed that increases in impervious surface are directly associated with decreased infiltration and recharge (Erickson & Stefan 2009). Urban environments are known to increase the quantity of stormwater and pollutant loading of adjacent surface water bodies (Cordery 1977; Göbel et al. 2007), and these impacts may intensify with respect to specific surface types, such as roadways (Harrison & Wilson 1985a, 1985b; Sansalone & Buchberger 1997; Ball et al. 1998).

Rainwater harvesting has been proposed as a solution to urban water problems as a form of basic infrastructure for potable water collection (Helmreich & Horn 2009), as a means of minimizing household consumption (Fewkes 1999; Villareal & Dixon 2005; Fletcher et al. 2007; Slyś 2009; Steffen et al. 2013), maintaining and preserving sewer infrastructure (Vaes & Berlamont 1999; Parkinson et al. 2005), and attenuating stormwater volume (Furumai 2008; Khastagir & Jayasuriya 2010; Burns et al. 2015). It has been suggested that the benefits of rainwater harvesting are synergistic, and that the attenuation of stormwater is a direct consequence of collection upstream (Angrill et al. 2012).

A considerable amount of rainwater harvesting research has been dedicated to the analysis of surfaces and the quality of runoff. Many have opted to analyze roof surface materials as a means of comparing pollutant profiles (Lee et al. 2012), including those pertaining to first-flush water (Gikas & Tsihrintzis 2012). Further comparisons of surface roughness, or rugosity, have been made to assess differences in runoff quantity (Farreny et al. 2011). Some have opted to focus on the quality of stormwater from streets (Cordery 1977; Ball et al. 1998; van Wesemael et al. 1998; Ramier et al. 2006; Göbel et al. 2007; Fletcher et al. 2008), and analyzing its suitability for lower quality uses such as toilet flushing (Nolde 2007). Kim et al. have studied the pollutant profiles of multiple urban surfaces, including roofs and roads (Kim et al. 2005).

Urban hydrology research often concerns the modelling of stormwater flows. Hydrological models have been proposed as a means of estimating urban runoff using geographic information system (GIS)-based, distributed hydrological modelling (Chormanski et al. 2008). Hydrological models have included residential rainwater harvesting tanks as a means of assessing reductions in stormwater quantity (Burns et al. 2012), and improvements in quality (Khastagir & Jayasuriya 2010). Comparisons of urban form have also been used to demonstrate impacts on stormwater quantity and quality (Grodek et al. 2011). It should be noted that the study of hydrological modelling is not fully understood (Mitchell et al. 2001), and that there is ‘significant variation in runoff volumes between simulated and experimental results’ (Ramier et al. 2006). It is also important to note that excessive harvesting of rainwater may be deleterious to the health of surrounding streams and water bodies (Fletcher et al. 2007).

GIS software has been used on a number of levels in the modelling of rainwater harvesting installations. Chiu et al. have used GIS to estimate optimal rainwater harvesting tank size. The authors consider the ‘water-energy nexus’ and recommend that the sizing of rainwater harvesting tanks should be based on water, energy and economic savings (Chiu et al. 2015). GIS has also been used to develop suitability maps for runoff harvesting and agricultural applications. In general, these maps combine data on rainfall, slope, soil type and land cover to identify suitable sites for rainwater harvesting installations, which can include check dams, terraces, and indigenous stormwater harvesting systems (de Winnaar et al. 2007; Jasrotia et al. 2009; Mahmoud & Alazba 2015). Some have developed decision support systems to identify suitable locations for rainwater harvesting infrastructure (Mbilinyi et al. 2007; Jha et al. 2014).

This study has been conducted as a preliminary exploration of rainwater harvesting and stormwater attenuation potential. It is intended to establish baseline stormwater estimates for Escambia County, Florida. The data resulting from this analysis can be used to reduce the local water footprint, and identify any modifications required for existing stormwater infrastructure. This is particularly important in Escambia County, Florida, which is subject to periodic and inordinately intense rainfall events.

MATERIALS AND METHODS

GIS

This study used GIS software (ESRI 2015), aerial photographs from the National Agriculture Imagery Program (NAIP 2010) and local airborne LiDAR data. It should be noted that there is a four-year lapse between the NAIP and LiDAR files, and that any development taking place during this period was not accounted for in the final analysis. Imagery files for the State of Florida are also accessible through the Land Area Boundary Information System (LABINS). However, the NAIP files were preferred for this analysis as they included a near infrared spectrum (Figure 1) which facilitated the calculation of Normalized Difference Vegetation Index (NDVI). NDVI is derived from the near infrared and visible red bands, as given by the following formula:  
formula
where VIS refers to the spectral reflectance of the visible regions, and NIR refers to the spectral reflectance of the near-infrared regions. NDVI is used to differentiate between areas of dense and sparse vegetation, water, buildings and roads. It is an indicator of ‘photosynthetic capacity’. In GIS, NDVI can be used to create a raster surface, with cell values between −1 and +1. For the purposes of this study, any area with an NDVI value above zero was classified as vegetation and therefore excluded from consideration as a possible impervious surface.
Figure 1

Sample of near infrared spectrum of NAIP imagery files.

Figure 1

Sample of near infrared spectrum of NAIP imagery files.

Airborne LiDAR data were used to make a high resolution (1 meter cell size) elevation surface of ground measurements. A subsequent surface was produced from all LiDAR points not classified as ‘ground’ measurements. This raster contains the elevation of non-ground surfaces such as trees, cars and rooftops. These surfaces were overlaid with the ‘ground’ values being subtracted from the ‘above ground’ cell values to identify pronounced differences in elevation between the two. The resulting surface was then reclassified to identify elevation differences of greater than 2.5 meters.

A third raster surface distinguishing between trees and rooftops was created to correct for false-positive classification errors resulting from forested areas that were cleared between 2006 and 2010. A surface roughness statistic was applied to a LiDAR surface, as the elevations of tree foliage are more irregular than the comparatively uniform and predictable surfaces of rooftops. Based on the standard deviation of the roughness statistic, each cell was reclassified as having one of two values (1 or 0), and in so doing, differentiating between trees or rooftops, as depicted in Figure 2.

Figure 2

Sample of LiDAR roughness and intensity map.

Figure 2

Sample of LiDAR roughness and intensity map.

The three surfaces were overlaid using the ‘Boolean AND’ function in GIS. First, the roughness and elevation surfaces were combined, and subsequently overlaid with the NDVI surface. The majority of rooftop surfaces were effectively identified in this manner, although a small number of isolated raster cells and irregularly shaped features were still present. In comparing the subsequent surface map with the original imagery files, these isolated cells were either misidentified as ‘rooftops’, or insufficient in size or thickness to constitute a viable rooftop catchment. These cells were removed by reclassifying the surface by region size and thickness. All areas less than 538 ft2 (50 m2) and 4.92 ft. (1.5 m) in thickness were excluded from the analysis. The remaining rooftop surfaces were expanded and contracted (1 meter) to make outlines more regular, as shown in Figure 3.

Figure 3

Outline of rooftop surfaces following the removal of non-viable catchment areas.

Figure 3

Outline of rooftop surfaces following the removal of non-viable catchment areas.

Assessment of water consumption

Household consumption data were acquired from the Emerald Coast Utilities Authority (ECUA) through a Freedom of Information Act (FOIA) request. There were 48,757 unique addresses listed in the ECUA data for 2015, 42,028 of which contained 12 months of consumption data. The remaining records accounted for only part of the 2015 billing cycle, and were therefore excluded from the analysis.

With household consumption data for 42,028 households, a simplified version of the water balance method was used to compare a given household's water consumption, with the rainwater harvesting potential of the household's roof. Only the existing roof was considered as a catchment area, disregarding the potential on any given site for constructing new catchment surfaces for water harvesting. Weather data were accessed from the National Oceanic and Atmospheric Administration (NOAA) National Centers for Environmental Information website. Daily precipitation totals were taken from the Pensacola Airport weather station (PENSACOLA 1.9 NE, FL, USA) for the years inclusive of 1950–2014. Average rainfall during this period was slightly less than 64 inches per year, with about 44% of rainfall occurring between May and September. As depicted in Figure 4, mean and median rainfall for each month has been calculated based on the daily precipitation totals given for the years 1950–2014. It has been observed that median values provide a more conservative estimate of rainfall totals (Texas Water Development Board 2005). All of the calculations in this study are based on median values.

Figure 4

Mean and median monthly rainfall totals for Pensacola, FL, 1950–2014.

Figure 4

Mean and median monthly rainfall totals for Pensacola, FL, 1950–2014.

It is important to note that many residential rainwater harvesting systems have common elements, depending on the end use of the harvested rainwater. Although these elements are not universal in design or size, systems generally require the following components: a catchment, gutters and downspouts, leaf guards, first flush diverters, a cistern or storage area, delivery systems, often including pumps, and a filtration system. The sizing of rainwater harvesting systems is primarily based on seasonal rainfall, the size of the collection surface, and building-specific demand. Theoretically, 0.62 gallons of rainwater can be harvested for every square foot of collection surface per inch of rainfall. In practice, 100% collection efficiency is not achievable due to evaporation, leaks in conveyance systems, splashing and overshoot of gutters and roof edges during intense rainfall events, and first flush requirements. The pitch and roughness of the collection surface should also be taken into account, as these factors also influence harvesting potential (Texas Water Development Board 2005).

Estimates of first flush vary considerably, depending on regional and material variables, slope and smoothness of the collection surface, the intensity of the rain event, the length of time between events, and the nature of the contaminants (Texas Water Development Board 2005). Farreny et al. have noted pronounced differences in rainwater harvesting potential between flat and sloped roofs, and surfaces with varied roughness coefficients (Farreny et al. 2011).

In an effort to provide an estimate of rainwater harvesting potential for a large area and number of samples (N = 42,028), this study assumed 85% collection efficiency for all roof surfaces due to the high percentage of asphalt shingles–upward of 90%–in the US residential roofing market, and the relative consistency of rainfall in Florida. In this study, seventy-five percent of rainfall events exceed 0.06 inches (1 mm), and this accounts for 99.7% of annual rainfall by volume. It is believed that the 85% collection efficiency is a reasonable estimate for potential collection.

Rainwater harvesting potential for each rooftop surface was compared with household consumption patterns (N = 42,028). Thus, the sample data were classified into two distinct groups: net positive households–those who consumed less than their rainwater harvesting potential – and net negative households – those who consumed more that their rainwater harvesting potential. In order to estimate maximum potential, unlimited storage capacity was assumed. Samples were compared for differences between means with independent t-tests using SPSS software (IBM 2013).

RESULTS

There are approximately 591,933,525 ft2 (54,992,424 m2) of roof surface in Escambia County, Florida, which represents approximately 23 billion gallons of potential stormwater runoff, given annual rainfall of approximately 64 inches. Of the household consumption data acquired from ECUA, 25,981 were classified as net positive and 16,047 as net negative. As such, 62% of households could meet their water demand through rainwater harvesting. The remaining 38% of households consumed more water than could potentially be harvested from their rooftop catchments, a sample of which is depicted in Figure 5.

Figure 5

Sample classification of net positive (blue) and net negative (orange) households.

Figure 5

Sample classification of net positive (blue) and net negative (orange) households.

The net positive and net negative samples were further analyzed for gallons of water consumed, area of rooftop catchment and parcel area or lot size using independent t-test analyses. Independent t-tests were used to identify significant differences between the sample populations and their respective means.

Mean water consumption for all 42,028 records was 80,679 gallons per year, with a standard deviation of 122,547 gallons. It is important to note that these data do not fit a normal distribution, although this is not uncommon with very large datasets. The differences between the net positive and net negative groups were analyzed. A mean of 46,044 gallons of water per year was used by the net positive group, with a standard deviation of 23,661. In contrast, a mean of 136,756 gallons of water was used by the net negative group, with a standard deviation of 182,594 as shown in Table 1. The results of the independent t-test analysis were determined to be significant (p < 0.001), with t= 62.61 (df = 16,379), and the 95% Confidence Interval (CI) = [87,873, 93,552], as shown in Table 2.

Table 1

Group statistics: comparison of means for gallons consumed, difference between net positive and net negative population group statistics

  Group statistics
 
 Mean Std. deviation Std. error mean 
Gallons consumed Net negative 16,047 136,756.3907 182,593.6473 1,441 
Net positive 25,981 46,043.8754 23,661.0771 146.8 
  Group statistics
 
 Mean Std. deviation Std. error mean 
Gallons consumed Net negative 16,047 136,756.3907 182,593.6473 1,441 
Net positive 25,981 46,043.8754 23,661.0771 146.8 
Table 2

Independent t-tests for differences between net positive and net negative population for gallons of water consumed

  Independent samples test
 
  Levene's Test for Equality of Variances
 
t test for Equality of Means
 
Sig. df Sig. (2-tailed) Mean difference Std. error difference 95% CI of the difference
 
Lower Upper 
Gallons consumed Equal variances assumed 1,523 0.000 79.011 42,026 0.000 90,713 1,148 88,462 92,963 
Equal variances not assumed   62.609 16,379 0.000 90,713 1,149 87,873 93,552 
  Independent samples test
 
  Levene's Test for Equality of Variances
 
t test for Equality of Means
 
Sig. df Sig. (2-tailed) Mean difference Std. error difference 95% CI of the difference
 
Lower Upper 
Gallons consumed Equal variances assumed 1,523 0.000 79.011 42,026 0.000 90,713 1,148 88,462 92,963 
Equal variances not assumed   62.609 16,379 0.000 90,713 1,149 87,873 93,552 

Mean roof area for all 42,028 records was 2,371 ft2 (220 m2), with a standard deviation of 786. Of the net positive records (N = 25,981), mean roof area was 2,485 ft2 (231 m2), with a standard deviation of 802. In contrast, mean roof area for the net negative records (N = 16,047) was 2,186 ft2 (203 m2), with a standard deviation of 720.6 (Table 3). A mean difference of −299.4 ft2 (75 m2) was observed, and determined to be significant (p < 0.001), with t = −39.62 (df = 36,716), with the 95% CI = [−314.2, −284.2], as shown in Table 4.

Table 3

Group statistics: comparison of means for roof area, difference between net positive and net negative population

 Group statistics
 
    Mean Std. deviation Std. error mean 
Roof area Net negative 16,047 2,186 720.6 5.689 
Net positive 25,981 2,485 802.0 4.976 
 Group statistics
 
    Mean Std. deviation Std. error mean 
Roof area Net negative 16,047 2,186 720.6 5.689 
Net positive 25,981 2,485 802.0 4.976 
Table 4

Independent t-tests for differences between net positive and net negative population for roof area

  Independent samples test
 
     
  Levene's Test for Equality of Variances
 
t-test for Equality of Means
 
t-test for Equality of Means
 
t-test for Equality of Means
 
Sig. df Sig. (2-tailed) Mean difference Std. error difference 95% CI of the difference
 
Lower Upper 
Roof area Equal variances assumed 114.5 0.000 −38.63 42,026 0.000 −299.4 7.750 −314.6 −284.2 
Equal variances not assumed   −39.62 36,716 0.000 −299.4 7.557 −314.2 −284.2 
  Independent samples test
 
     
  Levene's Test for Equality of Variances
 
t-test for Equality of Means
 
t-test for Equality of Means
 
t-test for Equality of Means
 
Sig. df Sig. (2-tailed) Mean difference Std. error difference 95% CI of the difference
 
Lower Upper 
Roof area Equal variances assumed 114.5 0.000 −38.63 42,026 0.000 −299.4 7.750 −314.6 −284.2 
Equal variances not assumed   −39.62 36,716 0.000 −299.4 7.557 −314.2 −284.2 

Mean parcel area was also calculated for all records (N = 42,028), and determined to be 20,294 ft2 (1,885 m2) with a standard deviation of 42,047. Again, an independent t-test was performed between the net positive and net negative groups. The normality assumption was not met, but variances were equal. The mean parcel area for the net positive group was 20,603 ft2 (1,914 m2), with a standard deviation of 38,162 (Table 5). The mean parcel area for the net negative group was 19,794 ft2 (1,839 m2), with a standard deviation of 47,668. The difference between means for parcel area were found to be insignificant, with p = 0.055 (n.s.) and t = −1.916 (df = 42,026), as shown in Table 6.

Table 5

Group statistics: comparison of means for parcel area, difference between net positive and net negative population

  Group statistics
 
  Mean Std. deviation Std. error mean 
Parcel area Net negative 16,047 19,794 47,668 376.3 
Net positive 25,981 20,603 38,162 236.8 
  Group statistics
 
  Mean Std. deviation Std. error mean 
Parcel area Net negative 16,047 19,794 47,668 376.3 
Net positive 25,981 20,603 38,162 236.8 
Table 6

Independent t-tests for differences between net positive and net negative population for parcel area

  Independent samples test
 
     
  Levene's Test for Equality of Variances
 
t-test for Equality of Means
 
t-test for Equality of Means
 
t-test for Equality of Means
 
Sig. df Sig. (2-tailed) Mean difference Std. error difference 95% CI of the difference
 
Lower Upper 
Parcel area Equal variances assumed 1.783 0.182 −1.916 42,026 0.055 −808.8 422.2 −1,636 18.601 
Equal variances not assumed   −1.819 28,506 0.069 −808.8 444.6 −1,680 62.58 
  Independent samples test
 
     
  Levene's Test for Equality of Variances
 
t-test for Equality of Means
 
t-test for Equality of Means
 
t-test for Equality of Means
 
Sig. df Sig. (2-tailed) Mean difference Std. error difference 95% CI of the difference
 
Lower Upper 
Parcel area Equal variances assumed 1.783 0.182 −1.916 42,026 0.055 −808.8 422.2 −1,636 18.601 
Equal variances not assumed   −1.819 28,506 0.069 −808.8 444.6 −1,680 62.58 

CONCLUSIONS

The analysis and estimation of impervious surfaces is an important first step toward understanding the hydrological dynamics of urban areas. Certainly, the application of the analysis presented in this paper goes beyond the simple estimation of rainwater harvesting potential. With respect to stormwater runoff, analysis of this type could be used to minimize the size of stormwater infrastructure, and identify ways of reducing impervious surfaces that generate high levels of pollution. It should also be noted that many impervious surfaces are underused for much of their lifetimes, such as parking lots. These surfaces are generally heat sinks, and consideration should be given to the effect on local climate and building energy consumption. The needs of natural systems should also be considered. The capture and use of rainwater may effectively reduce the water available for natural lymnological processes and local fauna.

This study was initiated for the purposes of estimating rainwater harvesting potential for residential structures in Escambia County, Florida. In comparing impervious, rooftop surface maps with household consumption data from the local utility (ECUA), it has been shown that 62% of local households could potentially meet all of their household water needs through rainwater harvesting. The remaining 38%, although unable to meet all household water demands, have the capacity to reduce household water consumption considerably.

There are significant differences between households that are capable of achieving net zero water consumption, and those that are not. There are significant differences in consumption patterns: a difference of 90,713 gallons between means. There are also significant differences in roof sizes: a difference of 299 ft2 between means, which equates to a difference of approximately 10,120 gallons in annual rainwater harvesting potential. Perhaps unexpectedly, there was no significant difference between the parcel sizes of the net positive and net negative households. It has been suggested that irrigation comprises 50% or more of household water consumption (Haley et al. 2007; Kibert 2008, p. 221). However, it should be noted that the ratio of indoor to outdoor water use varies considerably (Survis 2010), such that the use of irrigation may have more to do with plant selection and user behavior than considerations of parcel size. According to this study, there was no significant difference between the parcel sizes of the net positive and net negative samples.

DISCUSSION

It is fair to say that household water withdrawals could be decreased considerably through the implementation and codification of widespread rainwater harvesting infrastructure. In practice, the proposition of water consumption is much more complicated. Harvested rainwater is often deemed unfit for potable consumption by local and State health agencies (Ward et al. 2010; NCSL 2016), and there are important distinctions in water quality that should be observed with respect to end use. However, it is widely believed that laundry, toilet flushing and irrigation do not require potable quality water.

Policy makers should also examine the density of urban development, which may constrain the space available for water storage, and consequently, the amount of harvested rainwater water available per capita. This study assumed unlimited storage in residential buildings, which allows for the maximum volume available. This deviates considerably from typical rainwater harvesting practice. Future research should examine how harvested rainwater could benefit buildings in dense, urban environments. Similarly, considerations of specific components, space availability and cost would provide important insight into the refinement of impervious surface modeling and stormwater infrastructure.

ACKNOWLEDGEMENTS

We would like to thank Dr Nancy Bridier for her assistance with an unusually large dataset and the statistical analyses presented in this paper.

FUNDING

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

REFERENCES

REFERENCES
Angrill
,
S.
,
Farreny
,
R.
,
Gasol
,
C. M.
,
Gabarrell
,
X.
,
Viñolas
,
B.
,
Josa
,
A.
&
Rieradevall
,
J.
2012
Environmental analysis of rainwater harvesting infrastructures in diffuse and compact urban models of Mediterranean climate
.
International Journal of Life Cycle Assessment
17
,
25
42
.
Ball
,
J. E.
,
Jenks
,
R.
&
Aubourg
,
D.
1998
An assessment of the availability of pollutant constituents on road surfaces
.
The Science of the Total Environment
209
,
243
254
.
Bureau of Meteorology – Australian Government
2015
.
Burns
,
M. J.
,
Fletcher
,
T. D.
,
Walsh
,
C. J.
,
Ladson
,
A. R.
&
Hatt
,
B. E.
2012
Hydrologic shortcomings of conventional urban stormwater management and opportunities for reform
.
Landscape and Urban Planning
105
,
230
240
.
Burns
,
M. J.
,
Fletcher
,
T. D.
,
Duncan
,
H. P.
,
Hatt
,
B. E.
,
Ladson
,
A. R.
&
Walsh
,
C. J.
2015
The performance of rainwater tanks for stormwater retention and water supply at the household scale: an empirical study
.
Hydrological Processes
29
,
152
160
.
Chormanski
,
J.
,
Van de Voorde
,
T.
,
De Roeck
,
T.
,
Batelaan
,
O.
&
Canters
,
F.
2008
Improving distributed runoff prediction in urbanized catchments with remote sensing based estimates of impervious surface cover
.
Sensors
8
,
910
932
.
Cordery
,
I.
1977
Quality characteristics of urban storm water in Sydney, Australia
.
Water Resources Research
13
(
1
),
197
202
.
CSIRO & Bureau of Meteorology
2015
Climate Change in Australia Information for Australia's Natural Resource Management Regions: Technical Report
,
CSIRO and Bureau of Meteorology
,
Australia
.
de Winnaar
,
G.
,
Jewitt
,
G. P. W.
&
Horan
,
M.
2007
A GIS-based approach for identifying potential runoff harvesting sites in the Thukela River basin, South Africa
.
Physics and Chemistry of the Earth
32
,
1058
1067
.
Dougherty
,
M.
,
Dymond
,
R. L.
,
Grizzard
,
T. J.
,
Godrej
,
R. N.
,
Zipper
,
C. E.
&
Randolph
,
J.
2007
Quantifying long term hydrologic response in an urbanizing basin
.
Journal of Hydrologic Engineering
12
(
1
),
33
41
.
Erickson
,
T. O.
&
Stefan
,
H. G.
2009
Natural groundwater recharge response to urbanization: Vermillion River Watershed, Minnesota
.
Journal of Water Resources Planning and Management
135
,
512
520
.
ESRI
2015
ArcGIS Desktop: Release 10.1. Environmental Systems Research Institute, Redlands, CA
.
Farreny
,
R.
,
Morales-Pinzón
,
T.
,
Guisasola
,
A.
,
Taya
,
C.
,
Rieradevall
,
J.
&
Gabarrell
,
X.
2011
Roof selection for rainwater harvesting: quantity and quality assessments in Spain
.
Water Research
45
,
3245
3254
.
Fletcher
,
T. D.
,
Mitchell
,
V. G.
,
Deletic
,
A.
,
Ladson
,
T. R.
&
Séven
,
A.
2007
Is stormwater harvesting beneficial to urban waterway environmental flows?
Water Science and Technology
55
(
4
),
265
272
.
Fletcher
,
T. D.
,
Deletic
,
A.
,
Mitchell
,
V. G.
&
Hatt
,
B. E.
2008
Reuse of urban runoff in Australia: a review of recent advances and remaining challenges
.
Journal of Environmental Quality
37
,
S116
S127
.
Furumai
,
H.
2008
Rainwater and reclaimed wastewater for sustainable urban water use
.
Physics and Chemistry of the Earth
33
,
340
346
.
Gikas
,
G. D.
&
Tsihrintzis
,
V. A.
2012
Assessment of water quality of first-flush roof runoff and harvested rainwater
.
Journal of Hydrology
466–467
,
115
126
.
Göbel
,
P.
,
Dierkes
,
C.
&
Coldewey
,
W. G.
2007
Storm water runoff concentration matrix for urban areas
.
Journal of Contaminant Hydrology
91
,
26
42
.
Grant
,
S. B.
,
Fletcher
,
T. D.
,
Feldman
,
D.
,
Saphores
,
J. D.
,
Cook
,
P. L. M.
,
Stewardson
,
M.
,
Low
,
K.
,
Burry
,
K.
&
Hamilton
,
A. J.
2013
Adapting urban water systems to a changing climate: lessons from the millennium drought in Southeast Australia
.
Environmental Science and Technology
47
,
10727
10734
.
Grodek
,
T.
,
Lange
,
J.
,
Lekach
,
J.
&
Husary
,
S.
2011
Urban hydrology in mountainous middle eastern cities
.
Hydrology and Earth System Sciences
15
,
953
966
.
Haley
,
M. B.
,
Dukes
,
M. D.
&
Miller
,
G. L.
2007
Residential irrigation water use in Central Florida
.
Journal of Irrigation and Drainage Engineering
133
(
5
),
427
434
.
Harrison
,
R. M.
&
Wilson
,
S. J.
1985a
The chemical composition of highway drainage waters I: major ions and selected trace materials
.
The Science of the Total Environment
43
,
63
77
.
Helmreich
,
B.
&
Horn
,
H.
2009
Opportunities in rainwater harvesting
.
Desalination
248
,
118
124
.
IBM Corp. Released
2013
IBM SPSS Statistics for Windows, Version 22.0
.
IBM Corp.
,
Armonk, NY
.
IPCC
2014
Climate change 2014: synthesis report
. In:
Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change
(
Core Writing Team
,
Pachauri
,
R. K.
&
Meyer
,
L. A.
, eds).
IPCC
,
Geneva
,
Switzerland
, p.
151
.
Jha
,
M. K.
,
Chowdary
,
V. M.
,
Kulkarni
,
Y.
&
Mal
,
B. C.
2014
Rainwater harvesting planning using geospatial techniques and multi-criteria decision analysis
.
Resources, Conservation and Recycling
83
,
96
111
.
Khastagir
,
A.
&
Jayasuriya
,
L. N. N.
2010
Impacts of using rainwater tanks on stormwater harvesting and runoff quality
.
Water Science and Technology
62
(
2
),
324
329
.
Kibert
,
C. J.
2008
Sustainable Construction: Green Building Design and Delivery
,
2nd edn
.
John Wiley and Sons, Inc.
,
Hoboken, New Jersey
.
Kim
,
R.-H.
,
Lee
,
S.
,
Kim
,
Y.-M.
,
Lee
,
J. H.
,
Kim
,
S. K.
&
Kim
,
S.-G.
2005
Pollutants in rainwater runoff in Korea: their impacts on rainwater utilization
.
Environmental Technology
26
(
4
),
411
420
.
Lee
,
J. Y.
,
Bak
,
G.
&
Han
,
M.
2012
Quality of roof-harvested rainwater – comparison of different roofing materials
.
Environmental Pollution
162
,
422
429
.
Marvel
,
K.
&
Bonfils
,
C.
2013
Identifying external influences on global precipitation
.
Proceeding of the National Academy of Sciences of the United States
110
(
48
),
19301
19306
.
Mbilinyi
,
B. P.
,
Tumbo
,
S. D.
,
Mahoo
,
H. F.
&
Mkiramwinyi
,
F. O.
2007
GIS-based decision support system for identifying potential sites for rainwater harvesting
.
Physics and Chemistry of the Earth
32
,
1074
1081
.
Mitchell
,
V. G.
,
Mein
,
R. G.
&
McMahon
,
T. A.
2001
Modelling the urban water cycle
.
Environmental Modelling and Software
16
,
615
629
.
National Conference of State Legislatures (NCSL)
2016
State Rainwater Harvesting Laws and Legislation. Electronic Resource: http://www.ncsl.org/research/environment-and-natural-resources/rainwater-harvesting.aspx (accessed 13 September 2016)
.
National Agriculture Imagery Program (NAIP)
2010
Farm Service Agency, United States Department of Agriculture, Salt Lake City, UT
.
Parkinson
,
J.
,
Schütze
,
M.
&
Butler
,
D.
2005
Modelling the impacts of domestic water conservation on the sustainability of the urban sewerage system
.
Water and Environment Journal
19
(
1
),
49
56
.
Paul
,
M. J.
&
Meyer
,
J. L.
2001
Streams in the urban landscape
.
Annual Review of Ecology and Systematics
32
,
333
365
.
Ramier
,
D.
,
Berthier
,
E.
,
Dangla
,
P.
&
Andrieu
,
H.
2006
Study of the water budget of streets: experimentation and modelling
.
Water Science and Technology
54
(
6–7
),
41
48
.
Rippey
,
B.
2015
United States Drought Monitor. http://droughtmonitor.unl.edu/ (accessed 24 May 2015)
.
Sansalone
,
J. J.
&
Buchberger
,
S. G.
1997
Characterization of solid and metal element distributions in urban highway stormwater
.
Water Science and Technology
36
(
8–9
),
155
160
.
Scalenghe
,
R.
&
Marsan
,
F. A.
2009
The anthropogenic sealing of soils
.
Landscape and Urban Planning
90
,
1
10
.
Steffen
,
J.
,
Jensen
,
M.
,
Pomeroy
,
C. A.
&
Burian
,
S. J.
2013
Water supply and stormwater management benefits of residential rainwater harvesting in US cities
.
Journal of American Water Resources Association
49
(
4
),
810
824
.
Survis
,
F. D.
2010
Residential Lawn Water Use and Lawn Irrigation Practices: Wellington, Florida
.
Master Thesis
,
Florida Atlantic University
,
Boca Raton, Florida
,
USA
.
Texas Water Development Board
2005
The Texas Manual on Rainwater Harvesting
,
3rd edn.
Texas Water Development Board
,
Austin, Texas
.
United States Department of Agriculture (USDA), Natural Resources Conservation Service
2015
California Water Supply Outlook, May 10, 2015
.
United States Environmental Protection Agency (US EPA)
2015
Rate of Precipitation Change Indicator. http://www.epa.gov/climatechange/science/indicators/weather-climate/precipitation.html (accessed 25 May 2015)
.
Vaes
,
G.
&
Berlamont
,
J.
1999
The impact of rainwater reuse on CSO emissions
.
Water Science and Technology
39
(
5
),
57
64
.
van Wesemael
,
B.
,
Poesen
,
J.
,
Solé Benet
,
A.
,
Cara Barrionuevo
,
L.
&
Puidefábregas
,
J.
1998
Collection and storage of runoff from hillslopes in a semi-arid environment: geomorphic and hydrologic aspects of the aljibe system in Almeria Province, Spain
.
Journal of Arid Environments
40
,
1
14
.
Ward
,
S.
,
Memon
,
F. A.
&
Butler
,
D.
2010
Harvested rainwater quality: the importance of appropriate design
.
Environmental Science and Technology
61
(
7
),
1707
1714
.